Development of a Brief Computerized Adaptive Testing Version of the SCL-90: Validation of a Six-Factor Model Using a...

The Symptom Checklist-90 (SCL-90) is widely used in Chinese adolescents, but its practical utility is limited by its length and by structural problems in the traditional ten-factor model. To address these issues, this study developed and validated a brief computerized adaptive testing (CAT) system based on a theoretically informed six-factor model. Using a clinical sample of 10,052 adolescent outpatients, confirmatory factor analysis showed the traditional ten-factor model had a substantial identification problem: a non-positive definite matrix from extremely high factor correlations. In contrast, the proposed six-factor model had acceptable fit (CFI = 0.875, RMSEA = 0.061). The CAT system was developed via multidimensional item response theory. In post hoc simulations with a stopping rule (SE < 0.4), average test length was reduced from 90 to 37 items (58.9% reduction), maintaining high agreement with full-scale estimates (≥ 0.89) and acceptable marginal reliability (≥ 0.76). External validation with the SDS, SAS, and Y-BOCS provided preliminary support for external validity. Overall, this study supports a clinically meaningful six-factor representation of the SCL-90 and provides an efficient adaptive assessment framework for Chinese adolescents in clinical settings.